Question Answering on CRUD Single-hop (test)
0.3754BLEU-1Meta-chunker-1.5B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Meta-chunker-1.5BChunking Method=Meta-chunker, Backbone=Qwen2.5-1.5B2025.03 | 0.3754 | 0.276 | 0.4445 | |
| Qwen2.5-72BChunking Method=Direct LLM Inference, Backbone=Qwen2.5-72B2025.03 | 0.3721 | 0.2743 | 0.4405 | |
| Qwen2.5-14BChunking Method=Direct LLM Inference, Backbone=Qwen2.5-14B2025.03 | 0.365 | 0.2679 | 0.4351 | |
| Llama_indexChunking Method=LlamaIndex default2025.03 | 0.362 | 0.2682 | 0.4326 | |
| OriginalChunking Method=Rule-based/Original2025.03 | 0.3515 | 0.2548 | 0.4213 | |
| LumberChunkerChunking Method=Model-based (Qwen2.5-14B)2025.03 | 0.3456 | 0.2542 | 0.416 | |
| Semantic ChunkingChunking Method=Semantic-based2025.03 | 0.3382 | 0.2462 | 0.4131 |